setup-wizard-generator
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill is specifically designed to handle sensitive credentials (e.g., Stripe keys, GitHub secrets). However, it follows safe practices by using
ask_secret(hidden input) and instructing users to store them in.envfiles or GitHub secrets rather than hardcoding them into the script itself. - [COMMAND_EXECUTION]: The generated scripts use
gh secret setandgh variable setto interact with GitHub's CLI. It also uses utility commands likewslview,xdg-open, andopento launch browsers. These are standard administrative actions for the skill's stated purpose. - [INDIRECT_PROMPT_INJECTION]: The skill reads from local terminal files (e.g.,
.env,README,.env.example) to scope the procedure. If these files contain malicious instructions disguised as setup steps, the agent might generate a wizard that tricks the user into exfiltrating data or performing unsafe actions (Category 8c - Tool output poisoning). Assessment: LOW severity as it requires a local attacker or malicious repo and the execution is ultimately human-in-the-loop.
Audit Metadata